View Full Version : FPRIME windows 7 64 bit crashing
mav3rick
10-01-2010, 12:10 PM
Topic says all.. after some time in process of rendering fprime crashes. so if i have 100 frame sequence it could crash after 20 minute or something of rendering... there is no fixed pattern and happens randomly .
No meter if i start 64 bit or 32 bit version result is same....
Specs:
Win 7 64 bit
12 gb ram
980x
460 gtx 1 gb
digitaldoc
11-30-2010, 08:10 PM
I have the exact same problem. Seems to happen after 70-74 frames for me.
tajino
11-30-2010, 08:37 PM
Same here, I thought I was the only one having this problem, time to notify Steve.
Strange thing is if you click cancel in the program crash dialog, Fprime will continue rendering for another 50-80 frames before it actually crashes for good.
digitaldoc
11-30-2010, 08:53 PM
I didn't notice that, but I couldn't close lightwave after the crash, even from the task manager, just had to reboot. (Running two instances of fprime render simultaneouly)
tajino
11-30-2010, 10:39 PM
I didn't notice that, but I couldn't close lightwave after the crash, even from the task manager, just had to reboot. (Running two instances of fprime render simultaneouly)
When you dismissed the first LW program crash dialog (not the one from Windows7), Fprime will continue to render if you check the task manager you'll notice it is still working, which is the reason why you cannot end task it.
digitaldoc
11-30-2010, 11:39 PM
Aha. Thanks!
tajino
12-13-2010, 01:57 PM
Has anyone tried to contact Worley Labs support lately. There seems to be no reply after sending 3 emails regarding this Win7 render crash issue.
evolross
12-13-2010, 07:50 PM
Good luck! This is exactly why I'm excited about VPR. FPrime's technology was always amazing but the support and information about the product's direction has always been a mystery.
digitaldoc
12-13-2010, 08:08 PM
This is about rendering with FPrime as opposed to LW renderer. In my most recent HD project for a TV commercial, FPrime rendered at better quality at 2 min/frame as compared to 6 min/frame with similar AA results, a definite time-saver despite having to babysit the renders.
Just tried a 360 frame animation here, Windows 7 64, no crash here. Is there anything specific in the scene that you're using?
digitaldoc
12-13-2010, 09:49 PM
Mine was 9.6, 2 instances running simultaneously.
tajino
12-15-2010, 12:58 PM
Just tried a 360 frame animation here, Windows 7 64, no crash here. Is there anything specific in the scene that you're using?
Crash usually occurs with this kind of scene :
I'm using 2 bounce montecarlo radiosity to render a better looking pseudo ambient occlusion pass at level 40 Classic Non-Refinable, light quality 0.500
Scenes is about 400-500 frames, have to be relatively complex, 30 items or more, with subpatch geometry, total polycount about 400-600 K. Contains bones and displacement. About 30 surfaces in total, all surfaces are the same diffuse white for the pseudo AO pass.
Usually it will crash at about 150-200 frames or so.
Sensei
12-15-2010, 01:26 PM
Crash usually occurs with this kind of scene [...]
Then start turning things off (or somebody else who is experiencing this), one by one, rerendering after every modification, until it's rendering without crashing, and you will have the answer what is really causing it..
Random crash means usually that memory was damaged at some point, because of exceeding array with offset lower than 0 or higher than allocated area, or using freed memory. Crash happens while using damaged bytes.. So, it can be hours later than when it's physically damaged. That's why they're so hard to find.
Boris Goreta
12-23-2010, 08:13 AM
I had random crashes if radiosity was set higher then 100% so if that is the case try lowering it to 100%.
tajino
01-21-2011, 11:49 PM
Then start turning things off (or somebody else who is experiencing this), one by one, rerendering after every modification, until it's rendering without crashing, and you will have the answer what is really causing it..
Random crash means usually that memory was damaged at some point, because of exceeding array with offset lower than 0 or higher than allocated area, or using freed memory. Crash happens while using damaged bytes.. So, it can be hours later than when it's physically damaged. That's why they're so hard to find.
That is only if it crashes on all Windows OSes, but now it only crash on Win7 with scenes that work perfectly fine on Vista and XP.
Contacting Worley labs is of no help too, no reply after months. Don't know what's up with them.
Only workaround I can think of now is to create a batch file to use WSN and break up the scenes into 10-20 frames chunk and render it that way to avoid the crash that usually occurs after >50 continuous frames.
What else, LW uncached 2 bounce montecarlo radiosity is too slowwwww.... to even be practical, cached can't provide enough details. Kray QMC use together with frame interpolated cache is pretty good and fast comparable to fprime, but then it is too anal about the mesh and AA is poor.
ericsmith
01-22-2011, 12:22 PM
I was having crash problems on a new 6 core I7 machine I just got specifically for rendering. It currently has XP installed, and would crash after around 15 - 20 minutes of rendering. The same scene, running on XP on my 9650 quad core would not crash.
After some back and forth, I tried turning off hyperthreading on the new machine and the crashing has stopped completely.
I'm actually curious to see if this works on some of your setups, as I'm currently considering upgrading the new machine to windows 7 64 bit.
Eric
digitaldoc
02-17-2011, 03:05 PM
Just happened again. Doing a rehash of a project I did last Fall, few changes by the client.
I found that when the crash message appears if I just click to report it, then click on through, everything seems to be rendering fine, with no actual crash.
If I act like it has crashed and try to close the program, it freezes and computer needs a reboot.
(Actually running 3 instances of FPrime rendering full HD (58% of 16GB memory usage)
palpal
04-10-2011, 04:30 PM
I also have this bug and it really makes a pain for me... just want to leave it rendering... hate this... and I have tried so much to get rid of this bug but can not get it to be crash free...
Your PAL :help:
digitaldoc
04-10-2011, 05:04 PM
I also have this bug and it really makes a pain for me... just want to leave it rendering... hate this... and I have tried so much to get rid of this bug but can not get it to be crash free...
Your PAL :help:
Try to click the report it button, then click on through. It just keeps rendering as it is supposed to.
Good luck!
palpal
04-10-2011, 05:18 PM
Thanx... but will that bug then reapear and one must do this again... want my fprime to just work while I am not here if you know? he he... thanx again!
Wonder what that bug might come from though... can it be an afterwave from the last first subdevision order maybe(this were a huge bug in the early fprime days I can remember with pain...he he been out of animation for some years really... good to be back on it again)... try to set all mine to last!
Your PAL:bowdown:
digitaldoc
04-10-2011, 05:25 PM
It happens with XP, Vista and Windows 7.
With LW 9.6 32 and 64 bit, AND LW 10 32 and 64 bit.
It's probably not going to be fixed. Seems to be a trigger to the crash message without actually ending the program.
palpal
04-10-2011, 05:30 PM
That is a bit funny... monty python material that is... think about it hard for one moment... the report crash window creates a crash.... ha ha ha... how funny is that! Newtek... fix fix! :agree:
I think you are right... there must be something really strange here... the same thing actually happens if I render huge amount of frames wit montecarlo or final gathering with lw render, and not using cashe, something one do not want to do... but then LW also crashes... hmmm... maybe it really has something to do with ram handeling?
Your PAL :D
palpal
04-11-2011, 09:42 AM
http://www.maxi-pedia.com/3GB+switch+Windows+boot.ini+3+GB
http://www.pronetworks.org/forums/3gb-switch-as-option-to-vista-t100600.html
Can any of these be to help for someone here... I have not testet it out yet as I am not in my office right now!
Your PAL :beerchug:
tajino
04-18-2011, 09:18 AM
Looks like the workaround by turning off Hyperthreading as suggested by ericsmith did solved the crashing problem on my dual w5580 and x5650 so far.
Did anyone reported this problem to Worley labs, the more the better.
vBulletin® v3.8.2, Copyright ©2000-2012, Jelsoft Enterprises Ltd.